feat(helper): ✨ remove links, emails, and handles from descriptions

laxamentumtech · Aug 17, 2023 · e80819d · e80819d
1 parent 59a5d5c
commit e80819d
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 4 deletions.
diff --git a/src/helpers/authors/audible/ScrapeHelper.ts b/src/helpers/authors/audible/ScrapeHelper.ts
@@ -2,6 +2,7 @@ import * as cheerio from 'cheerio'
 import { htmlToText } from 'html-to-text'
 
 import { ApiAuthorOnBook, ApiAuthorProfile, ApiAuthorProfileSchema } from '#config/types'
+import cleanupDescription from '#helpers/utils/cleanupDescription'
 import fetch from '#helpers/utils/fetchPlus'
 import SharedHelper from '#helpers/utils/shared'
 import {
@@ -130,7 +131,7 @@ class ScrapeHelper {
 		}
 
 		// Description
-		const description = this.getDescription(dom)
+		const description = cleanupDescription(this.getDescription(dom))
 		// Genres
 		const genres = this.helper.collectGenres(
 			this.asin,

diff --git a/src/helpers/books/audible/ApiHelper.ts b/src/helpers/books/audible/ApiHelper.ts
@@ -16,6 +16,7 @@ import {
 	AudibleSeries,
 	AudibleSeriesSchema
 } from '#config/types'
+import cleanupDescription from '#helpers/utils/cleanupDescription'
 import fetch from '#helpers/utils/fetchPlus'
 import SharedHelper from '#helpers/utils/shared'
 import {
@@ -269,9 +270,11 @@ class ApiHelper {
 				}
 				return authorJson
 			}),
-			description: htmlToText(this.audibleResponse['merchandising_summary'], {
-				wordwrap: false
-			}).trim(),
+			description: cleanupDescription(
+				htmlToText(this.audibleResponse['merchandising_summary'], {
+					wordwrap: false
+				})
+			).trim(),
 			formatType: this.audibleResponse.format_type,
 			...(this.categories && {
 				genres: [...this.getGenres(), ...this.getTags()]

diff --git a/src/helpers/utils/cleanupDescription.ts b/src/helpers/utils/cleanupDescription.ts
@@ -0,0 +1,14 @@
+// regex to remove any emails
+const emailRegex = /([a-zA-Z0-9._-]+@[a-zA-Z0-9._-]+\.[a-zA-Z0-9_-]+)/g
+
+// Regex to remove any twitter handles
+const twitterRegex = /(^|[^@\w])@(\w{1,15})\b/g
+
+// Regex to remove any urls (with or without http(s))
+const urlRegex = /((https?:\/\/)?[\w-]+(\.[\w-]+)+\.?(:\d+)?(\/\S*)?)/gm
+
+const cleanupDescription = (description: string): string => {
+	return description.replace(emailRegex, '').replace(twitterRegex, '').replace(urlRegex, '').trim()
+}
+
+export default cleanupDescription